MSV Not Operating Properly on Mark V

We are attempting to restart our steam turbine after a loss of trip oil. We are having issues with the MSV2 valve "popping" open once the operators select reset. The controller is a Mark V TMR. Once operators select reset on the controller, the turbine starts to roll off. The highest we let it roll off to was a little over 360 RPMs before we opted to trip the unit. It was noticed because in our procedure it states that the V2 extraction control valve should open to 86%, but due to the pressure from the MSV2 leak by, the logic told the V2 to stay shut.

We recently came out of an outage where the valves were rebuilt. The start after the outage was successful and we did not experience the MSV2 valve opening when the controller was reset. This was the first trip since the work was done.

MSV2 is the valve that has the startup pilot.

Valve command when stroked/Position Feedback
-25/.43
-5/.43
-1/.8
0/1.45

This is the same thing that we saw when attempting to start up; only the valve command goes from -25 to 0.

We have changed the servo on the MSV2 three times, all with the same result.

We have re-calibrated our LVDTs and they are all reading accurately. I adjusted them to when fully close and hydraulic pressure off, the feedback position is 0.00. That position is true as long as the controller is sending a -25 command signal. However, when given 0 command, the feedback would show between 1.8 to 2.

I have attempted to change the null bias. I adjusted it all the way from 40 to 3 but did not notice any change that would be considered significant. On the MSV2 it is set at 22.49. I do not have a super clear understanding of when to change that valve and to what extent.

Our understanding of the servos was that they were failsafe. In the event of a loss of all three servos, the control valve would go shut. That was not our experience. We pulled the cannon plug connector to the servo and enabled our hydraulic oil trip circuit. The valve went fully open.

We tested to see what we would experience if we completed a servo polarity check with the servo reversed and with only one servo connected to the controller, the valve went fully open.

We could also not get the valve to shut when driving the valve with only one servo either. It would open fairly easily, but when given a close command it would stop around 9%.

We were also under the impression that if one of the servo polarities was reversed that the other two servos proper wired would overcome the one reversed servo. That was not our case. The valve would open and travel fairly well, but when give a 0 command, it would hang up around 9% open.

I am out of ideas. Any direction on what may be the cause or advice on troubleshooting is appreciated.
 
joe45678,

What Diagnostic Alarms are being annunciated by the Mark V?

>Valve command when stroked/Position Feedback
>-25/.43
>-5/.43
>-1/.8
>0/1.45
>
>This is the same thing that we saw when attempting to start
>up; only the valve command goes from -25 to 0.

<i>Based on the information provided,</i> I would say at least part of the problem is with the LVDT calibration, or, more specifically, that when the unit tripped and the MSV2 (Main Stop Valve #2?) slammed shut something happened to cause the LVDT mounting to change position.

Without being able to see the P&ID(s) for the trip oil/hydraulic oil it's really difficult to say what else might have happened. But, it certainly seems there is something amiss hydraulically, as well.

And, it doesn't really seem as if the Mark V is the cause of the problem. The servo-valve output seems to be changing (without knowing what Diagnostic Alarms are being annunciated), and it would appear that the trip solenoid output(s) are also working correctly.

I'm not very good with steam turbines without P&IDs, but I want to try to address some of the issues raised with the comments below. And all of my comments refer to bipolar electro-hydraulic servo-valves (such as those manufactured by Moog and Abex) as used on most GE-design turbine applications (heavy duty gas turbines and combined-cycle steam turbines, as well as many industrial steam turbines (such as used at many paper mills)).

>We have changed the servo on the MSV2 three times, all with
>the same result.

Electro-hydraulic servo-valves are perceived to be mythical, magical devices. They are really very simple devices that convert an electrical signal (or in the case of TMR applications) three signals into hydraulic oil flow. They are very similar to I/P (current-to-pressure) positioners on many pneumatic control valves, except the servos are bipolar and the position loop is closed in the controller instead of at the device. In other words, instead of 4-20 mA for 0-100% stroke, the servo-valve stops hydraulic flow to/from the hydraulic actuator when the current is, effectively, 0 mA (plus the null bias--more on that below). When the Mark V servo output current goes negative on most GE turbine applications then the valve should start to open. The more negative the current the faster the valve should move. When the current goes back to "zero" the hydraulic flow will go back to zero and the valve will stop at its current position. When the Mark V servo output current goes positive on most GE-design turbine applications then the valve should start to close. The more positive the current being applied the faster the valve will move closed.

Now, null bias current. For most of the electro-hydraulic servo-valves used on GE-design turbines there is a single spring in the servo-valve assembly which is always applying pressure to a movable spool piece (which controls the flow-rate and direction of hydraulic oil through the servo-valve) to try to move the device being actuated (a control valve, usually) to the position that shuts off the flow of steam (or fuel or air) to the turbine. This is ALWAYS happening, all the time. The spring tension can't be switched on or off, it's always acting on the spool piece to try to shut off the flow of steam (or fuel or air).

In the absence of all electric signals to the servo-valve (the three servo-valve output currents from the Mark V), the function of the spring in the servo-valve is move the spool piece in the servo-valve to the position that causes hydraulic flow to/from the device being controlled to reduce the flow of steam (or fuel or air) to the turbine. That's the fail-safe bit--in the absence of any electric signal the spring in the servo-valve (often called the fail-safe spring, and sometimes, mistakenly, called the null bias spring) mechanically moves the spool piece in the servo-valve to the position that shuts off the flow of steam (or fuel or air) to the turbine.

Now, because there is this continual spring tension trying to move the spool piece to shut off the flow of steam (or fuel or air) to the turbine the servo-valve output currents from the turbine control system must ALWAYS be applying a small amount of current to overcome the spring tension in order to be able to shut off the flow of hydraulic oil through the servo-valve in order to be able to stop the control valve in a particular position. This current is called null bias current--the null position is the position of the spool piece in the servo-valve that shuts off the flow of hydraulic oil to or from the hydraulic actuator which keeps the device being controlled (a control valve) in a particular position. The null bias current required for MOST electro-hydraulic servo-valves used on GE-design turbine applications is specified to be -0.8 mA, +/-0.4 mA, and is set by the servo-valve manufacturer using a small adjusting screw on the side of the servo-valve on a test bench using a specific hydraulic pressure and flow-rate. When the total of the three servo-currents being applied by the turbine control system equal -0.8 mA, +/-0.4 mA, then the spool piece in the servo-valve is in the position that shuts off the flow of hydraulic oil to/from the hydraulic actuator of the device being controlled, and that means the device being controlled will stop in its present position, say, 14.57%, or 83.49%, or approximately whatever the reference is from the control processor(s) in the turbine control system.

It's important to know that the null bias current CANNOT be properly adjusted in the field with the little screw on the side of the servo-valve. Yes; some sites/people have said they have successfully adjusted the null bias current with the little screw, but, that is questionable because the entire circumstances of the situation are not known. And, without knowing the flow-rate of hydraulic oil through the servo-valve, it's virtually impossible to say for certain that the adjustment was "successful."

In the TMR control system, the reference from all three control processors should be identical (under ideal conditions), so the null bias current is set to be one-third of the required (specified) amount of null bias current for the device. So, one-third of -0.8 mA is -0.2667 mA, and since in the Mark V (and Mark VI and Mark VIe) 10 mA is equal to 100%, -0.2667 mA equal -2.667%. And, in the I/O Configurator, the negative signal is automatically applied to the value in the Null Bias Current field, so when you see 2.667 it actually is -2.667, percent.

Now, for how the servo-valve uses the three signals. The servo-valve "sums" the values of the three currents being applied to the servo-valve's three coils, so, if <R>'s servo current is -3.50% and <S>'s servo current is -1.17% and <T>'s servo current is -2.63%, then amount of current being applied is (-3.5 + -1.17 + -2.63 =) -7.3%, or -0.73mA, which is within the -0.8 mA, +/-0.4 mA range, and if the device position reference was 43.54% <i>if the LVDT feedback was calibrated properly</i> the actual device position should be approximately 43.54%, plus-or-minus a couple of tenths of a percent (position).

Let's say that one of the control processors though the turbine should be tripped, it's output would be very negative, say, -40%, or -4 mA, and that current would be summed with the other two processor's in the servo-valve--which would have to be putting out enough current to overcome that negative value to keep the device open to keep the flow of steam (or fuel or air) flowing to the turbine. And, it should be noted that the actual position of the device, as well as the LVDT feedback from the device, would be slightly less than the reference position--because one processor was trying (very hard!) to shut off the flow of steam (or fuel or air) while the other two would have to be working (very hard!) to overcome that action, and the result would be a position (actual and indicated) slightly less than reference.

Now, when to adjust the null bias current value? <b>ONLY AFTER IT HAS BEEN VERIFIED THAT THE LVDT POSITION IS NEARLY EXACTLY EQUAL TO THE ACTUAL DEVICE POSITION</b> (in other words the calibration of the LVDT has been verified against the actual position of the device and found to be nearly exactly equal to actual device position) but the actual position (and therefore the actual position) is not very close to the reference position <b>THEN AND ONLY THEN</b> should the null bias current be changed. And, it if it has to be changed to anything more than 4 or anything less than 1.33 (-4% or -1.33%, in reality), then there is a problem with the servo-valve and it should be replaced. And, the amount of change of the null bias current value can't be calculated using any formula--it's kind of a seat-of-the-pants process.

Now, if I read all of your information correctly and I presume there are no relevant Diagnostic Alarms, it would seem that there is something wrong with the valve re-build. If you're rotating the position of the servo-valve on the manifold block and you are seeing the same results in both positions, then I would say it's NOT the servo or the Mark V. And, if you are changing the null bias value, downloading the new value(s) to the three control processors, and then re-booting the three processors one at a time to make the changes active and you aren't seeing some change, then there is something mechanically or hydraulically wrong with the valve/actuator and not with the Mark V.

>We have re-calibrated our LVDTs and they are all reading
>accurately. I adjusted them to when fully close and
>hydraulic pressure off, the feedback position is 0.00. That
>position is true as long as the controller is sending a -25
>command signal. However, when given 0 command, the feedback
>would show between 1.8 to 2.

I would say there is something wrong with calibration, because at 0% reference (with no relevant Diagnostic Alarms), if the LVDT feedback matches the actual valve position then the feedback (and the actual valve position) should be 0%. And, if not, then <i><b>starting with the null bias set at 2.67 for each of the three control processor</i></b> one can try adjusting the null bias (in this case, I would make the null bias less than 2.67) to make the actual (and indicated) position equal to a reference of 0.

>I have attempted to change the null bias. I adjusted it all
>the way from 40 to 3 but did not notice any change that
>would be considered significant. On the MSV2 it is set at
>22.49. I do not have a super clear understanding of when to
>change that valve and to what extent.

WOW!!! 22.49. That's just an insane value--for most any electro-hydraulic servo-valve used on an GE-design turbine. Something's definitely wrong.

>We tested to see what we would experience if we completed a
>servo polarity check with the servo reversed and with only
>one servo connected to the controller, the valve went fully
>open.

>We could also not get the valve to shut when driving the
>valve with only one servo either. It would open fairly
>easily, but when given a close command it would stop around
>9%.

This should NOT happen with a properly functioning valve- and valve actuator on any typical GE-design application employing a servo-valve. There are SOME GE applications which use non-typical servos and applications, and without knowing more about the application and based on some of your previous posts it's presumed this is a typical application on a GE combined-cycle steam turbine using typical servos. (Again, I'm not terribly experienced with steam turbine applications, but I am very experienced with Mark V/VI/VIe, and this doesn't make any sense for typical GE designs.)

Again, I don't believe any of the above issues are being caused by the Mark V--or even the servo(s).

Something happened to the valve actuator and/or the valve when the unit tripped. Something is not right with the valve/valve actuator. Changing the servos and getting the same results is a good indication of that. Repeated unexpected behaviours under varying conditions is also a good indication of that.

Presuming there are no relevant Diagnostic Alarms, and there weren't any BEFORE this trip event (after the rebuild), then even though there are LOTS of wires and LEDs and chips and displays and electrons and mystery inside the Mark V panel--the problem is most likely NOT the result of anything in the Mark V. (That 22.49 null bias value is VERY suspect, however--VERY!)

Please write back to let us know what you find.
 
I don't think we had any relevant diagnostic alarms present. I will check on that.

I recalibrated the LVDTs the valve position is showing about +0.10% at true 0.

We tested the theory that the servo current should go to a negative value to allow for the valve to open. That was not the case with our MSV2. The MSV2_CUR signal stayed + for the entire valve stroke. Only once it got to fully open did the signal go negative.

The previous statement along with the fact that when we tested the "failsafe" functionality of the servo, it failed. The valve went fully open instead of closed.

We consulted with our mechanical engineer and the servo manufacturer and believe that we have had the wrong servo installed since startup. We think that the spring is on the wrong side of the servo. This seems to be supported by what we have found. We are going to continue to pursue that route until we come to a conclusion.

We made some changes to the interpolator block for MSV2 and we are not seeing the valve come off seat when we reset the turbine. This is not a fix for what we believe the problem to be. At the best case, this could be a patch to allow for us to restart the turbine. The steam/water in our plant is the priority over electricity.

We have adjusted the null bias to a value less than 2.67 and it seems to hold the valve closer to setpoint. The value found in the ioconfig of 22.49 is more than likely from persons in the years past trying to make the valve work. The servo manufacturer supported your statements of the null bias setting and adjustment.

I am still improving my knowledge of the hydraulic circuits, but I noticed you made a statement about changing the orientation of the servo on the manifold block. If we did, what could we expect? I am still going over it, but there is a pilot port in on of the corners of the servo block that has to line up, would that prevent it from operating?

Thank your for the support, I will share what we figure out with having the wrong servos installed.
 
joe45678,

>I don't think we had any relevant diagnostic alarms
>present. I will check on that.

Diagnostic Alarms are not nuisances (some may be, but they can be resolved in every case), and can help point to an issue with a card or input/output "channel." People overlook and ignore Diagnostic Alarms at their peril.

>I am still improving my knowledge of the hydraulic circuits,
>but I noticed you made a statement about changing the
>orientation of the servo on the manifold block. If we did,
>what could we expect? I am still going over it, but there is
>a pilot port in on of the corners of the servo block that
>has to line up, would that prevent it from operating?

I apparently misread the statement below from the original post:

>We tested to see what we would experience if we completed a
>servo polarity check with the servo reversed and with only
>one servo connected to the controller, the valve went fully
>open.

I have seen people inadvertently--and purposely--change the orientation of the servo on the hydraulic manifold "block" when troubleshooting. Are you certain the servo is oriented properly on the block to begin with?

As for what would happen if you reversed/changed the orientation, it's impossible to say without seeing the P&ID with the manifold block on it.

The ALIP block's parameters may also have been part of the problem--and it would seem that there are multiple issues contributing to your experience/problem. Most likely, as you say, people have been mucking with various control system parameters in an attempt to try to work around what is a serious and fundamental issue.

A servo-valve (current) polarity check should ALWAYS be performed any time a servo-valve is replaced. It <b>IS NOT</b> necessary to "re-calibrate" the device when a servo is replaced--calibration does nothing to affect the servo or regulator control. Calibration <b>ONLY</b> affects LVDT feedback scaling--<b>NOTHING ELSE.</b> (This is one of the many myths and wives' tales (falsehoods) about servos.) One is not calibrating the servo, or the regulator (the software element on the TCQA/TCQC card that drives the servo output current)--one is only performing a scaling of LVDT feedback. And, to do a calibration properly, one needs to know exactly where the device with the LVDTs is physically at--either using some built-in or attached scale or dial indicator (or in the case of IGVs a machinist's protractor or similar measuring instrument) to compare to the scaled LVDT feedback. And, since LVDTs are no different than pressure switches or temperature sensors which are routinely "calibrated," one needs to record the as-found condition of the calibration and then, and only then, re-calibrate if necessary. (Of course, if one replaces an LVDT, it must be calibrated. And, if a device equipped with LVDTs is disassembled for maintenance or refurbishment and the LVDTs are removed, then the LVDT feedback must also be re-calibrated before re-starting the unit. But, as a general rule, LVDT feedback calibration (scaling) does NOT drift (unless the LVDT armature and/or core loosen during operation) and does not require "calibration." LVDTs are extensively used in aircraft and rockets and other similar devices, simply because they are highly accurate (when stably mounted, mechanically) and do not experience frequent drift. They are proven technology, except in the mythology and folklore of turbine control.)

And, because servos and LVDTs so often are found together, and since they are highly misunderstood (thanks to the dearth of documentation by the designer/packager of the turbines and turbine control systems) and frequently (improperly) blamed for many ills and problems for which they are not responsible. (Oil quality is the most common cause of servo failures and improper operation--far and away the most common cause. Oil quality, and formulation, is key to long servo service and proper operation. Many servos, and the servo manufacturer, are (improperly) blamed for poor servo life and operation, when the problem can be directly attributed to poor oil quality and recent changes in oil formulations. Even the OEM knows this, but is reluctant to admit to it. The second most common cause of servo "failure" is improper servo (current) polarity, caused by no check of polarity at the time the servo is installed. New servos out of the box have been found to have reversed coil polarity even though the color coding of the coil leads is identical to the one being replaced. Refurbished servos also have known to have the same problem.)

PLEASE do keep us updated on the outcome and findings of this investigation!!!
 
We confirmed with the turbine manufacturer that we indeed had been using the wrong servo since the plant was commissioned.

The servo manufacturer who we had been communicating with directly, without confirmation from the turbine manufacturer, was skeptical about selling us the one that we felt we needed. He told us that there were literally thousands of the model of servos we had been using on the market. The model of servo we felt that we needed, there were eight.

The MSV2 valve only regulates until the turbine reaches partial arc. Once it makes it to partial arc, the MSV2 valve is basically out of the picture until a trip or shutdown. I guess everyone else in the past just accepted the battle of fighting it instead of fixing it.
 
joe45678,

Thank you very much for the feedback--and good on you for following through on this with the servo- and turbine manufacturers.

It would be great if you would share with us the turbine manufacturer and the model of turbine, as well as the servo manufacturer. It seems both have been helpful in your troubleshooting, and that's good to know in the future.

I want to take the opportunity to point out that the Mark V was NOT the problem, and that parameters in the Mark V had apparently been "jiggered" to get around the real issue.

Thanks again for the feedback, and if you could share additional information it could also be helpful for others. One last thing I want to point out is that owners can and should contact manufacturers when they have issues with components and equipment--even if the issue has been ongoing for many years. Sometimes one gets very lucky and gets the right person who can help with their issue, even if it's just a parameter change or a suggestion for more optimal operation, and as in this case help with determining the proper servo for the application.
 
The steam turbine manufacturer is General Electric. The turbine is an SAC2, automatic extraction condensing steam turbine.

The servo manufacturer is MOOG.

There were other parties who contributed to helping resolve the issue. PSM and MD&A provided support troubleshooting to get the issue resolved. MD&A supported the mechanical and control portions because they had rebuilt the steam valves. PSM supported troubleshooting the control system and hydraulic circuits.

All parties were offsite and communicating by phone and email.
 
joe45678,

Thank you very much for the information, and for the continuing feedback during the resolution of this problem!

I hope the unit has been returned to service and that everything checked out during the re-start and that any past problems are just that now--a thing of the past.
 
Top